Exploratory Data Analysis

by John W. Tukey

The approach in this introductory book is that of informal study of the data. Methods range from plotting picture-drawing techniques to rather elaborate numerical summaries. Several of the methods are the original creations of the author, and all can be carried out either with pencil or aided by hand-held calculator.

Genres: Science, Mathematics, Nonfiction, Business, Reference, Self Help, Textbooks

712 pages, Paperback

First published January 11, 1977

About the author

John Wilder Tukey (1915–2000) was an American mathematician best known for development of the FFT algorithm and box plot. The Tukey range test, the Tukey lambda distribution, the Tukey test of additivity, and the Teichmüller–Tukey lemma all bear his name.

Reviews

[Michael Scott · Rating 5 out of 5]

TODO full review:
+ This is a classic book everyone in science should read.
+ I have traversed this book since somewhere in the middle of my Ph.D., until, finally, a decade later, I consider I've read it. But the point is not to read it entirely, only to see and wonder at the description of some of the plot types you have surely used (surely the box and whiskers, currently so used it's even just the box plot? how about the box plot over time?), at some of the amazing ideas crammed into this book (the need for exploratory data analysis, a precursor of all the data-driven techniques we now use), at the prescient words: Our relationship with the computer [...] Much of what we have learned to do to data can be done by hand [...] LONG BEFORE one can find a computing system--to say nothing of getting the data entered. It will be a long time until this fails to be so. (p.663).
+ So many new designs! Stem and leaf (plus many cousins), box and whiskers (plus relatives), two-and-a-half-dimensional plots, two-/three-/multivariate plots, smoothing, annotating, etc.

[Max Lybbert · Rating 4 out of 5]

I read this book after discovering exploratory data analysis from the NIST/SEMATECH e-Handbook of Statistical Methods (available online). It's obvious that Tukey was a master at gaining understanding from batches of numbers. Unfortunately, this book did not make me a master. To be fair, I skipped the chapter exercises. Perhaps a little experience actually applying the subject matter would give me more confidence.

I'm impressed that the charts in this book were drawn by hand. Occasionally Tukey would provide advice on the mechanics of drafting or the mechanics of manually sorting large batches of numbers. Even in a world with cheap computers willing to do the drudge work, this advice is occasionally relevant.

Unfortunately, it's hard to determine who would enjoy reading this book. People wanting to apply exploratory data analysis are probably better off with the NIST/Sematech handbook (the preface to this book includes the warning, "Many of the techniques are less than ten years old in their present form--some will approve noticeably. And where a technique is very good, it is not at all certain that we yet know why it is"). People wanting to learn statistics are probably better off studying a more mainstream book, simply because Tukey will teach vocabulary and approaches that aren't widely used. Nobody will understand a report that refers to c'rank, froots, cuts, hinges, or several of the smoothing techniques presented.

However, I'm glad to have read this book, and I will read Data Analysis and Regression: A Second Course in Statistics. Perhaps this book is best recommended to people who want a deeper understanding of batches of numbers, even if the result won't be distributed to others.

[Mason · Rating 3 out of 5]

it has some neat ideas, but it's outdated. there's too much about techniques for graphing by hand

[Lecture Freed · Rating 4 out of 5]

Tukey provides a unique view to exploratory data analysis that to my knowledge has been lost. He provides a literal hands on approach to the topic of data analysis. In my opinion it is still a great read even though his methods of analysis are a bit dated. The key take away from this book are the principles for exploratory data analysis that Tukey points out. The exercises should be used as means to refine ones understanding of these ideas and can be either completed by hand or with some programming knowledge solved with a computer.

[Mrunal]

JHHJDGHGJHBKJNCKJHJHGUGJH

[Carlos Rendon · Rating 3 out of 5]

A fascinating look into the not so distant past of data analysis and visualization. Including computation by hand. Most of the material is no longer relevant, with a computer you can look at all the data almost instantly.

[Данило Судин · Rating 4 out of 5]

Джон Тьюкі - справжній класик і один з творців сучасного аналізу даних (варто хоча б згадати про boxplot чи метод множинних порівнянь Тьюкі в дисперсійному аналізі). Проте ця книга вже доволі застаріла. В принципі, сам автор в останньому розділі пише про перспективи її застарівання: потужні комп’ютери з’являться нескоро, а тому більшість обчислень треба буде робити вручну. Станом на 1977 р. Джон Тьюкі був правий, але у 2015 р. більшість обчислень немає сенсу проводити вручну: це довше і менш точно. Тому перші 7 глав книги є цікавими і цінними, але решта тексту - вже морально застаріла.

Втім, це важлива книга для розуміння розвідкового аналізу даних. І тому оминати її при вивченні статистичного аналізу даних не варто.